"""UnstructuredDocumentParserのテスト"""
import pytest
from unittest.mock import Mock, patch
from datetime import datetime
from src.adapter.unstructured_document_parser import UnstructuredDocumentParser
from src.domain.document import Document
from src.domain.chunk import Chunk
class TestUnstructuredDocumentParser:
"""UnstructuredDocumentParserのテストクラス"""
def test_parse_PDFファイルを解析した場合_適切なElementリストが返されること(self):
# ------------------------------
# 準備 (Arrange)
# ------------------------------
parser = UnstructuredDocumentParser()
file_path = "/docs/sample.pdf"
# Unstructuredライブラリのモック
mock_elements = [
Mock(text="First paragraph text", metadata={"page_number": 1}),
Mock(text="Second paragraph text", metadata={"page_number": 1}),
Mock(text="Third paragraph text", metadata={"page_number": 2}),
]
# ------------------------------
# 実行 (Act)
# ------------------------------
with patch(
"src.adapter.unstructured_document_parser.partition"
) as mock_partition:
mock_partition.return_value = mock_elements
elements = parser.parse(file_path)
# ------------------------------
# 検証 (Assert)
# ------------------------------
assert len(elements) == 3
assert elements[0].text == "First paragraph text"
assert elements[1].text == "Second paragraph text"
assert elements[2].text == "Third paragraph text"
mock_partition.assert_called_once_with(filename=file_path)
def test_parse_空のファイルを解析した場合_空のリストが返されること(self):
# ------------------------------
# 準備 (Arrange)
# ------------------------------
parser = UnstructuredDocumentParser()
file_path = "/docs/empty.txt"
mock_elements = []
# ------------------------------
# 実行 (Act)
# ------------------------------
with patch(
"src.adapter.unstructured_document_parser.partition"
) as mock_partition:
mock_partition.return_value = mock_elements
elements = parser.parse(file_path)
# ------------------------------
# 検証 (Assert)
# ------------------------------
assert len(elements) == 0
mock_partition.assert_called_once_with(filename=file_path)
def test_parse_ファイルが存在しない場合_例外が発生すること(self):
# ------------------------------
# 準備 (Arrange)
# ------------------------------
parser = UnstructuredDocumentParser()
file_path = "/docs/nonexistent.pdf"
# ------------------------------
# 実行 (Act) & 検証 (Assert)
# ------------------------------
with patch(
"src.adapter.unstructured_document_parser.partition"
) as mock_partition:
mock_partition.side_effect = FileNotFoundError("File not found")
with pytest.raises(FileNotFoundError):
parser.parse(file_path)
def test_split_elements_適切なサイズのElementsを分割した場合_適切なChunkリストが返されること(
self,
):
# ------------------------------
# 準備 (Arrange)
# ------------------------------
parser = UnstructuredDocumentParser()
document = Document(
file_path="/docs/test.txt",
file_name="test.txt",
file_type="txt",
content="test content",
file_last_modified=datetime(2025, 6, 13, 10, 0, 0),
relative_path="test.txt",
)
mock_elements = [
Mock(text="Short text 1", metadata={"page_number": 1}),
Mock(text="Short text 2", metadata={"page_number": 1}),
Mock(text="Another short text 3", metadata={"page_number": 2}),
]
# chunk_elementsのモック（1つのチャンクに結合される想定）
mock_chunked_elements = [
Mock(
text="Short text 1 Short text 2 Another short text 3",
metadata={"page_number": 1},
)
]
# ------------------------------
# 実行 (Act)
# ------------------------------
with patch(
"src.adapter.unstructured_document_parser.chunk_by_title"
) as mock_chunk_elements:
mock_chunk_elements.return_value = mock_chunked_elements
chunks = parser.split_elements(mock_elements, document)
# ------------------------------
# 検証 (Assert)
# ------------------------------
assert len(chunks) == 1 # 1つのチャンクが作成される
for chunk in chunks:
assert isinstance(chunk, Chunk)
assert chunk.source_document == document
assert len(chunk.text) > 0
assert len(chunk.text) <= parser.max_characters
def test_split_elements_長いElementを分割した場合_適切なチャンク数が作成されること(
self,
):
# ------------------------------
# 準備 (Arrange)
# ------------------------------
parser = UnstructuredDocumentParser(max_characters=100) # 小さなサイズでテスト
document = Document(
file_path="/docs/long.txt",
file_name="long.txt",
file_type="txt",
content="long content",
file_last_modified=datetime(2025, 6, 13, 10, 0, 0),
relative_path="long.txt",
)
# 長いテキストのElement（モック）
long_text = "A" * 250 # max_charactersの2.5倍の長さ
mock_elements = [Mock(text=long_text, metadata={"page_number": 1})]
# chunk_elementsのモック（複数のチャンクに分割される想定）
mock_chunked_elements = [
Mock(text="A" * 100, metadata={"page_number": 1}),
Mock(text="A" * 100, metadata={"page_number": 1}),
Mock(text="A" * 50, metadata={"page_number": 1}),
]
# ------------------------------
# 実行 (Act)
# ------------------------------
with patch(
"src.adapter.unstructured_document_parser.chunk_by_title"
) as mock_chunk_elements:
mock_chunk_elements.return_value = mock_chunked_elements
chunks = parser.split_elements(mock_elements, document)
# ------------------------------
# 検証 (Assert)
# ------------------------------
assert len(chunks) == 3 # 3つのチャンクに分割される
for chunk in chunks:
assert len(chunk.text) <= parser.max_characters
def test_split_elements_空のElementsを指定した場合_空のリストが返されること(self):
# ------------------------------
# 準備 (Arrange)
# ------------------------------
parser = UnstructuredDocumentParser()
document = Document(
file_path="/docs/empty.txt",
file_name="empty.txt",
file_type="txt",
content="empty",
file_last_modified=datetime(2025, 6, 13, 10, 0, 0),
relative_path="empty.txt",
)
mock_elements = []
# ------------------------------
# 実行 (Act)
# ------------------------------
chunks = parser.split_elements(mock_elements, document)
# ------------------------------
# 検証 (Assert)
# ------------------------------
assert len(chunks) == 0
def test_split_elements_異なるディレクトリの同名ファイル_ChunkIDが重複しないこと(
self,
):
# ------------------------------
# 準備 (Arrange)
# ------------------------------
parser = UnstructuredDocumentParser()
# 異なるディレクトリの同名ファイルを模擬
document1 = Document(
file_path="/project/dir1/readme.txt",
file_name="readme.txt",
file_type="txt",
content="content 1",
file_last_modified=datetime(2025, 6, 13, 10, 0, 0),
relative_path="dir1/readme.txt",
)
document2 = Document(
file_path="/project/dir2/readme.txt",
file_name="readme.txt",
file_type="txt",
content="content 2",
file_last_modified=datetime(2025, 6, 13, 10, 0, 0),
relative_path="dir2/readme.txt",
)
# モックElements
from unstructured.documents.elements import Text
# 実際のUnstructured Elementを使用
mock_element1 = Text(text="This is content from dir1")
mock_element2 = Text(text="This is content from dir2")
# ------------------------------
# 実行 (Act)
# ------------------------------
chunks1 = parser.split_elements([mock_element1], document1)
chunks2 = parser.split_elements([mock_element2], document2)
# ------------------------------
# 検証 (Assert)
# ------------------------------
assert len(chunks1) == 1
assert len(chunks2) == 1
chunk1 = chunks1[0]
chunk2 = chunks2[0]
# Chunk IDが異なることを確認
assert chunk1.id == "dir1/readme.txt_chunk_0"
assert chunk2.id == "dir2/readme.txt_chunk_0"
assert chunk1.id != chunk2.id
# エピソード名も異なることを確認
from src.domain.group_id import GroupId
group_id = GroupId("test")
episode1 = chunk1.to_episode(group_id)
episode2 = chunk2.to_episode(group_id)
assert episode1.name == "dir1/readme.txt - chunk_0"
assert episode2.name == "dir2/readme.txt - chunk_0"
assert episode1.name != episode2.name
assert episode1.source_description == "Source file: dir1/readme.txt"
assert episode2.source_description == "Source file: dir2/readme.txt"
assert episode1.source_description != episode2.source_description
